Development of Sirna Payloads to Target KRAS -Mutant Cancer
Total Page:16
File Type:pdf, Size:1020Kb
Published OnlineFirst August 6, 2014; DOI: 10.1158/2159-8290.CD-13-0900 RESEARCH ARTICLE Development of siRNA Payloads to Target KRAS -Mutant Cancer Tina L. Yuan 1 , Christof Fellmann 2 , Chih-Shia Lee 3 , Cayde D. Ritchie 1 , Vishal Thapar 2,4 , Liam C. Lee 3 , Dennis J. Hsu 3 , Danielle Grace 2,4 , Joseph O. Carver 3 , Johannes Zuber 2,5 , Ji Luo 3 , Frank McCormick 1 , and Scott W. Lowe 2,4,6 Downloaded from cancerdiscovery.aacrjournals.org on October 4, 2021. © 2014 American Association for Cancer Research. Published OnlineFirst August 6, 2014; DOI: 10.1158/2159-8290.CD-13-0900 ABSTRACT RNAi is a powerful tool for target identifi cation and can lead to novel therapies for pharmacologically intractable targets such as KRAS. RNAi therapy must com- bine potent siRNA payloads with reliable in vivo delivery for effi cient target inhibition. We used a func- tional “Sensor” assay to establish a library of potent siRNAs against RAS pathway genes and to show that they effi ciently suppress their targets at low dose. This reduces off-target effects and enables combination gene knockdown. We administered Sensor siRNAs in vitro and in vivo and validated the delivery of KRAS siRNA alone and siRNA targeting the complete RAF effector node (A/B/CRAF) as promising strategies to treat KRAS -mutant colorectal cancer. We further demonstrate that improved therapeutic effi cacy is achieved by formulating siRNA payloads that combine both single-gene siRNA and node-targeted siRNAs (KRAS + PIK3CA/B ). The customizable nature of Sensor siRNA payloads offers a universal platform for the combination target identifi cation and development of RNAi therapeutics. SIGNIFICANCE: To advance RNAi therapy for KRAS -mutant cancer, we developed a validated siRNA library against RAS pathway genes that enables combination gene silencing. Using an in vivo model for real-time siRNA delivery tracking, we show that siRNA-mediated inhibition of KRAS as well as RAF or PI3K combinations can impair KRAS -mutant colorectal cancer in xenograft models. Cancer Discov; 4(10); 1182–97. ©2014 AACR. INTRODUCTION RNAi provides an alternative therapeutic approach to small- molecule and antibody-based therapeutics for inhibiting gene Mutations in the RAS family of small GTPases, particularly function. RNAi can, in principle, be applied to reversibly silence KRAS, occur in 30% of all human cancers and are often asso- any target gene (reviewed in refs. 8 and 9 ), thereby increasing ciated with resistance to chemotherapy and targeted therapy. the druggable landscape from 10% to virtually 100% of the Somatic mutations lock RAS in the GTP-bound state, leading genome ( 10, 11 ). Because all siRNAs bear structural similarity to constitutive activation of its downstream effector path- and are likely to have comparable pharmacokinetic profi les, ways. Despite intense efforts, pharmacologic inhibition of their use as therapeutics would facilitate drug formulation, KRAS itself ( 1 ) and inhibition of individual effector kinases preclinical testing, and development of combination therapies. downstream of KRAS such as RAF ( 2, 3 ) and MEK ( 4–6 ) have Although delivery is still a major challenge, more reliable lipid so far been unsuccessful in treating KRAS -mutant tumors. and polymeric nanoparticles are in development to deliver Combinations of MEK and PI3K inhibitors are currently siRNA payloads to target tissues (reviewed in ref. 12 ). Regard- being evaluated in clinical trials, although toxicity in normal less, treatment effi cacy would benefi t from the development tissue could limit their therapeutic window ( 7 ). Thus, effec- of optimized siRNA payloads that potently and specifi cally tive and targeted treatment against KRAS-driven cancer is an silence well-validated target genes at low dose. urgent and unmet clinical need. Beyond its potential as a therapeutic modality, RNAi is a useful tool for identifying and validating new drug targets. This has proved particularly powerful in cancer research, where 1 Helen Diller Family Comprehensive Cancer Center, University of Cali- fornia, San Francisco, San Francisco, California. 2 Cold Spring Harbor shRNA or siRNA screens have been used to identify genes that Laboratory, Cold Spring Harbor, New York. 3 Laboratory of Cancer Biol- are selectively required for the proliferation and survival of ogy and Genetics, Center for Cancer Research, National Cancer Institute, cancer cells. However, the identifi cation and current in silico 4 Bethesda, Maryland. Memorial Sloan Kettering Cancer Center, New York, prediction of effective shRNAs and siRNAs remains imprecise, New York. 5 Research Institute of Molecular Pathology, Vienna, Austria. 6 Howard Hughes Medical Institute, New York, New York. resulting in low success rates. Consequently, screening high- order shRNA/siRNA combinations is not possible without Note: Supplementary data for this article are available at Cancer Discovery fi rst establishing a collection of functionally validated RNAi Online (http://cancerdiscovery.aacrjournals.org/). triggers. In addition, off-target effects, which can be due to T.L. Yuan, C. Fellmann, and C.-S. Lee contributed equally to this article. sequence-dependent and sequence-independent gene deregu- Current address for C. Fellmann: Mirimus, Inc., Cold Spring Harbor, New lation (reviewed in ref. 13 ), must be minimized for meaningful York. interpretation of phenotypic outcomes. Corresponding Authors: Scott W. Lowe, Memorial Sloan Kettering Cancer To overcome these limitations, we used a previously Center, 415 East 68th Street, Z-1114, New York, NY 10065. Phone: 646- described “Sensor” assay ( 14 ) to generate a functionally 888-3342; Fax: 646-888-3347; E-mail: [email protected] ; Frank McCor- validated library of RNAi sequences against RAS path- mick, [email protected] ; and Ji Luo, [email protected] . way genes. We show that Sensor siRNAs effi ciently ablate doi: 10.1158/2159-8290.CD-13-0900 their gene targets at low nanomolar concentrations in vitro , © 2014 American Association for Cancer Research. which decreased off-target effects and enabled the use of OCTOBER 2014CANCER DISCOVERY | 1183 Downloaded from cancerdiscovery.aacrjournals.org on October 4, 2021. © 2014 American Association for Cancer Research. Published OnlineFirst August 6, 2014; DOI: 10.1158/2159-8290.CD-13-0900 RESEARCH ARTICLE Yuan et al. high-order siRNA combinations to codeplete multiple correlation between the initial algorithmic rank and the rank genes. By applying Sensor siRNAs in vitro and in vivo , we obtained through the biologic Sensor assay ( Fig. 1E ), show- identifi ed single-gene and combination-gene payloads that ing that prediction tools alone are still insuffi cient to select inhibit the growth of KRAS -mutant colorectal cancer. Thus, the most potent shRNAs (see also Supplementary Fig. S1C the use of Sensor siRNAs for target discovery and develop- and S1D). Conversely, the Sensor assay identifi ed potent ment of customized siRNA payloads can lead to nanoparticle- shRNAs for nearly every gene, with top-ranked shRNAs show- based treatments for KRAS -mutant cancer and provide a ing similar scores across all genes ( Fig. 1F and Supplementary blueprint for similar strategies to target other key nodes in Table S2). More than 91% of the top fi ve human (344 of cancer maintenance. 375) and >88% of the top fi ve mouse (333 of 375) shRNAs scored >3 (Supplementary Fig. S1C and S1D), a threshold RESULTS score defi ned by positive-control shRNAs. This indicates that the preselection process of 65 shRNAs per gene provided A Functionally Validated RNAi Library Targeting suffi cient coverage to identify several very potent shRNAs KRAS Pathway Genes per gene. To identify potent shRNAs targeting the RAS network, we applied the Sensor assay to evaluate candidate shRNA Sensor siRNAs Are Potent and On-Target sequences targeting 75 human genes and their mouse Synthetic shRNAs and siRNAs enter the endogenous orthologs ( Fig. 1A ). These genes encode many classes of microRNA pathway at different stages, but ultimately use proteins, including kinases, GTPases, and transcription fac- the same conserved machinery to downregulate their target tors. The Sensor assay interrogates large numbers of shRNAs genes. We thus hypothesized that potent shRNA sequences under conditions of single genomic integration (“single- could be directly converted into potent siRNA triggers. On copy”) for their ability to repress a cognate target sequence the basis of top-scoring Sensor shRNA sequences, we gener- placed downstream of a fl uorescent reporter expressed in cis ated corresponding 22mer Sensor siRNAs targeting human (14 ). Although we have previously established this assay and KRAS, KSR1, ARAF, BRAF, RAF1/CRAF, MAP2K1, MAP2K2, shown its potential to identify very potent shRNA sequences MAPK1, MAPK3, PIK3CA, PIK3CB, PIK3CD, AKT1, AKT2, and through full gene tiling, here we applied it to evaluate the effi - AKT3 . We then measured the knockdown effi ciency of three ciency of preselected candidate sequences targeting a much top Sensor shRNAs and two corresponding siRNAs target- larger set of genes to establish a functionally validated RNAi ing endogenous KRAS , and found that Sensor siRNAs retain library. To assemble the initial candidates, 65 shRNAs per >80% mRNA knockdown when transfected at concentrations gene were selected using a combination of bioinformatics as low as 0.5 nmol/L ( Fig. 2A and 2B and Supplementary predictions ( 15 ) and “Sensor rules” requiring shRNA-specifi c Fig. S2A). We extended this validation to ARAF , BRAF , and features